View source for 18463778. METHOD AND SYSTEM FOR A TEMPERATURE-RESILIENT NEURAL NETWORK TRAINING MODEL simplified abstract (Arizona Board of Regents on Behalf of Arizona State University)

=METHOD AND SYSTEM FOR A TEMPERATURE-RESILIENT NEURAL NETWORK TRAINING MODEL=


==Organization Name==

[[:Category:Arizona Board of Regents on Behalf of Arizona State University|Arizona Board of Regents on Behalf of Arizona State University]]


[[Category:Arizona Board of Regents on Behalf of Arizona State University]]

==Inventor(s)==

[[:Category:Jae-sun Seo of Tempe AZ (US)|Jae-sun Seo of Tempe AZ (US)]][[Category:Jae-sun Seo of Tempe AZ (US)]]

[[:Category:Jian Meng of Tempe AZ (US)|Jian Meng of Tempe AZ (US)]][[Category:Jian Meng of Tempe AZ (US)]]

[[:Category:Li Yang of Tempe AZ (US)|Li Yang of Tempe AZ (US)]][[Category:Li Yang of Tempe AZ (US)]]

[[:Category:Deliang Fan of Tempe AZ (US)|Deliang Fan of Tempe AZ (US)]][[Category:Deliang Fan of Tempe AZ (US)]]

==METHOD AND SYSTEM FOR A TEMPERATURE-RESILIENT NEURAL NETWORK TRAINING MODEL - A simplified explanation of the abstract==


This abstract first appeared for US patent application 18463778 titled 'METHOD AND SYSTEM FOR A TEMPERATURE-RESILIENT NEURAL NETWORK TRAINING MODEL

==Simplified Explanation==

The method described in the abstract involves increasing the temperature-resiliency of a neural network by loading a neural network model into a resistive nonvolatile in-memory-computing chip, training the model using a progressive knowledge distillation algorithm, and utilizing a batch normalization adaptation algorithm.

* Loading a neural network model into a resistive nonvolatile in-memory-computing chip
* Training the model using a progressive knowledge distillation algorithm
* Injecting low-temperature noise values into the model and changing them to high-temperature noises
* Training the model using a batch normalization adaptation algorithm

== Potential Applications ==
The technology could be applied in industries where neural networks are used in high-temperature environments, such as aerospace, automotive, and industrial automation.

== Problems Solved ==
This technology addresses the issue of neural networks being sensitive to temperature variations, which can affect their performance and reliability in real-world applications.

== Benefits ==
The method increases the temperature-resiliency of neural networks, making them more robust and reliable in harsh environmental conditions.

== Potential Commercial Applications ==
The technology could be commercialized for use in manufacturing processes, autonomous vehicles, robotics, and other applications where neural networks need to operate in extreme temperature conditions.

== Possible Prior Art ==
Prior art in the field of neural network optimization and temperature resilience may include research papers, patents, or products that focus on improving the performance of neural networks in challenging environments.

=== Unanswered Questions ===

=== How does this method compare to existing techniques for improving the temperature-resiliency of neural networks? ===
This article does not provide a direct comparison with existing techniques or technologies in the field. It would be beneficial to understand the specific advantages and limitations of this method compared to other approaches.

=== What are the specific temperature ranges and conditions under which this method has been tested and proven effective? ===
The abstract does not mention the specific temperature ranges or environmental conditions in which the neural network model was trained and tested. Knowing this information would help assess the practical applicability of the technology in real-world scenarios.


==Original Abstract Submitted==

A method for increasing the temperature-resiliency of a neural network, the method comprising loading a neural network model into a resistive nonvolatile in-memory-computing chip, training the deep neural network model using a progressive knowledge distillation algorithm as a function of a teacher model, the algorithm comprising injecting, using a clean model as the teacher model, low-temperature noise values into a student model and changing, now using the student model as the teacher model, the low-temperature noises to high-temperature noises, and training the deep neural network model using a batch normalization adaptation algorithm, wherein the batch normalization adaptation algorithm includes training a plurality of batch normalization parameters with respect to a plurality of thermal variations.



[[Category:G06N3/08]]
[[Category:G06N3/0495]]